"State-space average models" are used on power conversion
simulations, for speed (accuracy, in the details, is on you).
I would be more inclined to a "stimulus-response" challenge
in transient analysis, looking for supply, ground and load
injection-points, perturbation and recovery from application-
realistic "stimuli".
Chopper mixing might produce some interesting frequency
domain artifacts and what this does to stability analysis
(especially "canned" stability analyses that just cough up
an intercept-point, may be fooled by spurii) that probably
wants inspection before belief.